fix: [Bug] Pipeline parameters (ParameterInteger, ParameterString) fail in ModelTrain (5504)

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-core/src/sagemaker/core/modules/utils.py

@@ -24,6 +24,7 @@
 
 from sagemaker.core.shapes import Unassigned
 from sagemaker.core.modules import logger
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
 
 
 def _is_valid_s3_uri(path: str, path_type: Literal["File", "Directory", "Any"] = "Any") -> bool:
@@ -129,19 +130,24 @@ def safe_serialize(data):
 
  This function handles the following cases:
  1. If `data` is a string, it returns the string as-is without wrapping in quotes.
- 2. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
+ 2. If `data` is of type `PipelineVariable`, it returns the PipelineVariable object
+ as-is for pipeline serialization.
+ 3. If `data` is serializable (e.g., a dictionary, list, int, float), it returns
  the JSON-encoded string using `json.dumps()`.
-3. If `data` cannot be serialized (e.g., a custom object), it returns the string
+4. If `data` cannot be serialized (e.g., a custom object), it returns the string
  representation of the data using `str(data)`.
 
  Args:
  data (Any): The data to serialize.
 
  Returns:
- str: The serialized JSON-compatible string or the string representation of the input.
+ str | PipelineVariable: The serialized JSON-compatible string, the string
+ representation of the input, or the PipelineVariable object as-is.
  """
  if isinstance(data, str):
  return data
+ elif isinstance(data, PipelineVariable):
+ return data
  try:
  return json.dumps(data)
  except TypeError:

sagemaker-core/tests/unit/core/modules/test_utils_safe_serialize.py

@@ -0,0 +1,82 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize in sagemaker.core.modules.utils with PipelineVariable support.
+
+Verifies that safe_serialize correctly handles PipelineVariable objects
+(e.g., ParameterInteger, ParameterString) by returning them as-is rather
+than attempting str() conversion which would raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.core.modules.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"

sagemaker-train/tests/unit/train/test_model_trainer_pipeline_variable.py

@@ -26,7 +26,7 @@
 
 from sagemaker.core.helper.session_helper import Session
 from sagemaker.core.helper.pipeline_variable import PipelineVariable, StrPipeVar
-from sagemaker.core.workflow.parameters import ParameterString
+from sagemaker.core.workflow.parameters import ParameterString, ParameterInteger
 from sagemaker.train.model_trainer import ModelTrainer, Mode
 from sagemaker.train.configs import (
  Compute,
@@ -176,3 +176,68 @@ def test_training_image_rejects_invalid_type(self):
  stopping_condition=DEFAULT_STOPPING,
  output_data_config=DEFAULT_OUTPUT,
  )
+
+
+class TestModelTrainerPipelineVariableHyperparameters:
+ """Test that PipelineVariable objects work correctly in ModelTrainer hyperparameters."""
+
+ def test_hyperparameters_with_parameter_integer(self):
+ """ParameterInteger in hyperparameters should be preserved through _create_training_job_args.
+
+ This test documents the exact bug scenario from GH#5504: safe_serialize
+ would fall back to str(data) for PipelineVariable objects, but
+ PipelineVariable.__str__ intentionally raises TypeError.
+ Before the fix, this call would have raised TypeError.
+ """
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"max_depth": max_depth},
+ )
+ # This call would have raised TypeError before the fix (GH#5504)
+ args = trainer._create_training_job_args()
+ # PipelineVariable should be preserved as-is, not stringified
+ assert args["hyper_parameters"]["max_depth"] is max_depth
+
+ def test_hyperparameters_with_parameter_string(self):
+ """ParameterString in hyperparameters should be preserved."""
+ algo = ParameterString(name="Algorithm", default_value="xgboost")
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={"algorithm": algo},
+ )
+ args = trainer._create_training_job_args()
+ assert args["hyper_parameters"]["algorithm"] is algo
+
+ def test_hyperparameters_with_mixed_pipeline_and_static_values(self):
+ """Mixed PipelineVariable and static values should both be handled correctly."""
+ max_depth = ParameterInteger(name="MaxDepth", default_value=5)
+ trainer = ModelTrainer(
+ training_image=DEFAULT_IMAGE,
+ role=DEFAULT_ROLE,
+ compute=DEFAULT_COMPUTE,
+ stopping_condition=DEFAULT_STOPPING,
+ output_data_config=DEFAULT_OUTPUT,
+ hyperparameters={
+ "max_depth": max_depth,
+ "eta": 0.1,
+ "objective": "binary:logistic",
+ "num_round": 100,
+ },
+ )
+ args = trainer._create_training_job_args()
+ hp = args["hyper_parameters"]
+ # PipelineVariable preserved as-is
+ assert hp["max_depth"] is max_depth
+ # Static values serialized to strings
+ assert hp["eta"] == "0.1"
+ assert hp["objective"] == "binary:logistic"
+ assert hp["num_round"] == "100"

sagemaker-train/tests/unit/train/test_safe_serialize.py

@@ -0,0 +1,92 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License"). You
+# may not use this file except in compliance with the License. A copy of
+# the License is located at
+#
+# http://aws.amazon.com/apache2.0/
+#
+# or in the "license" file accompanying this file. This file is
+# distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF
+# ANY KIND, either express or implied. See the License for the specific
+# language governing permissions and limitations under the License.
+"""Tests for safe_serialize with PipelineVariable support.
+
+Verifies that safe_serialize in sagemaker.train.utils correctly handles
+PipelineVariable objects (e.g., ParameterInteger, ParameterString) by
+returning them as-is rather than attempting str() conversion which would
+raise TypeError.
+
+See: https://github.com/aws/sagemaker-python-sdk/issues/5504
+"""
+from __future__ import annotations
+
+import pytest
+
+from sagemaker.train.utils import safe_serialize
+from sagemaker.core.helper.pipeline_variable import PipelineVariable
+from sagemaker.core.workflow.parameters import ParameterInteger, ParameterString
+
+
+class TestSafeSerializeWithPipelineVariables:
+ """Test safe_serialize handles PipelineVariable objects correctly."""
+
+ @pytest.mark.parametrize("param", [
+ ParameterInteger(name="MaxDepth", default_value=5),
+ ParameterString(name="Algorithm", default_value="xgboost"),
+ ])
+ def test_safe_serialize_returns_pipeline_variable_as_is(self, param):
+ """PipelineVariable objects should be returned as-is (identity preserved)."""
+ result = safe_serialize(param)
+ assert result is param
+ assert isinstance(result, PipelineVariable)
+
+ def test_pipeline_variable_str_raises_type_error(self):
+ """Confirm PipelineVariable.__str__ raises TypeError (the root cause of the bug)."""
+ param = ParameterInteger(name="TestParam", default_value=10)
+ with pytest.raises(TypeError):
+ str(param)
+
+
+class TestSafeSerializeBasicTypes:
+ """Regression tests: verify basic types still work after PipelineVariable support."""
+
+ def test_safe_serialize_with_string(self):
+ """Strings should be returned as-is without JSON wrapping."""
+ assert safe_serialize("hello") == "hello"
+ assert safe_serialize("12345") == "12345"
+
+ def test_safe_serialize_with_int(self):
+ """Integers should be JSON-serialized to string."""
+ assert safe_serialize(42) == "42"
+
+ def test_safe_serialize_with_float(self):
+ """Floats should be JSON-serialized to string."""
+ assert safe_serialize(3.14) == "3.14"
+
+ def test_safe_serialize_with_dict(self):
+ """Dicts should be JSON-serialized."""
+ result = safe_serialize({"key": "val"})
+ assert result == '{"key": "val"}'
+
+ def test_safe_serialize_with_bool(self):
+ """Booleans should be JSON-serialized."""
+ assert safe_serialize(True) == "true"
+ assert safe_serialize(False) == "false"
+
+ def test_safe_serialize_with_none(self):
+ """None should be JSON-serialized to 'null'."""
+ assert safe_serialize(None) == "null"
+
+ def test_safe_serialize_with_list(self):
+ """Lists should be JSON-serialized."""
+ assert safe_serialize([1, 2, 3]) == "[1, 2, 3]"
+
+ def test_safe_serialize_with_custom_object(self):
+ """Custom objects should fall back to str()."""
+
+ class CustomObj:
+ def __str__(self):
+ return "custom"
+
+ assert safe_serialize(CustomObj()) == "custom"